A smart home-oriented user behavior privacy protection method

Abstract

The application relates to the technical field of privacy protection, in particular to a user behavior privacy protection method for smart home, which is realized through transmission of a sending end and a receiving end, and specifically, a sensing link quality module and a dynamic flow confusion technology module are arranged in a transmission layer in the OSI five-layer model, and a demixing module is arranged in the transmission layer of the receiving end; application data of the sending end is transmitted to the transmission layer, the sensing link quality module feeds back current network link quality to the dynamic flow confusion technology module by monitoring the transmission state of TCP; the dynamic flow confusion technology module selects appropriate flow confusion technology and confusion strength according to the feedback, carries out flow confusion on the application data, and then encapsulates the application data into a data packet and transmits the data packet to a network layer; the demixing module is used to demix the received data packet in the transmission layer of the receiving end, and the original data before confusion is restored, so that the application layer can obtain accurate application data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of privacy protection, in particular to a user behavior privacy protection method for smart home. BACKGROUND

[0002] The application of the Internet of Things is becoming more and more common around the world. With the development of the Internet of Things, the number of smart home devices is increasing rapidly. According to the research and prediction of relevant agencies, by 2025, 21.3% of households worldwide will use smart home devices, and the total number of smart home devices will reach 5.44 billion.

[0003] Through connection with the Internet, smart home devices can realize remote control of intelligent terminals, automatic adjustment, and interaction with other devices, providing people with convenient and intelligent home life experience. However, the network traffic characteristics of these devices are closely related to user behavior, which can reflect the user's personal preferences, daily routines, etc. When device traffic is maliciously collected and analyzed, the user's privacy information is at risk of being misused, leaked, or used for commercial gain.

[0004] In recent years, for smart home traffic analysis, existing methods mostly use single obfuscation techniques (such as packet padding, traffic shaping, and false traffic injection) to reduce user behavior recognition rate. These methods can reduce the recognition rate to below 50%, but usually bring significant communication overhead, especially the increase of transmission delay, thereby affecting the quality of user experience. For low delay tolerant devices such as cameras, smart doorbells, and voice assistants, this delay problem is particularly evident. The delay can cause the loss of real-time performance of video monitoring and voice interaction, affecting the effectiveness of security monitoring and user satisfaction with smart devices. SUMMARY

[0005] The purpose of the present application is to provide a user behavior privacy protection method for smart home, aiming to solve the problem of significant communication overhead brought by traditional methods, thereby affecting the quality of user experience, and to achieve a balance between transmission efficiency and user behavior recognition rate.

[0006] To achieve the above purpose, the present application provides a user behavior privacy protection method for smart home, comprising the following steps:

[0007] Step 1: deploying a link quality perception module and a dynamic adjustment traffic obfuscation technology module in the transmission layer of the Open Systems Interconnection five-layer model at the sending end, and deploying a de-obfuscation module in the transmission layer at the receiving end;

[0008] Step 2: the sending end sends application data through the application layer, the link quality perception module monitors the transmission status of the Transmission Control Protocol, and feeds back the current network link quality to the dynamic adjustment traffic obfuscation technology module;

[0009] Step 3: The dynamic traffic obfuscation technology module selects the traffic obfuscation technology and obfuscation intensity based on the feedback, performs traffic obfuscation on the application data, and then encapsulates it into a data packet and transmits it to the network layer;

[0010] Step 4: The receiving end obtains the data packet from the network layer, deobfuscates it through the deobfuscation module, and restores the original data before obfuscation.

[0011] Optionally, the execution process of step 2 includes the following steps:

[0012] Step 2.1: The sender sends application data to the transport layer through the application layer;

[0013] Step 2.2: Monitor key parameters of the transport layer TCP connection;

[0014] Step 2.3: Conduct network link quality assessment;

[0015] Step 2.4: Perform anomaly detection to identify network anomalies.

[0016] Optionally, the key parameters of the transport layer transmission control protocol connection in step 2 include the sending window size, slow start threshold and round-trip time, and the link quality perception module evaluates the network link quality by analyzing the round-trip time, congestion status, delay and bottleneck.

[0017] Optionally, the traffic obfuscation technology in step 3 includes packet padding, packet segmentation, and false traffic injection. Packet padding determines the obfuscation strength by the small / large padding length range, packet segmentation determines the obfuscation strength by the simple / complexity, and false traffic injection determines the obfuscation strength by the low / high injection frequency.

[0018] Optionally, the packet padding mechanism is as follows:

[0019] Let M be the maximum transmission unit MTU, L be the original length of data packet P, Φ be the number of bytes to be added to data packet P, and take a random value between 1 and β;

[0020] Ф=random_int(1,β)

[0021] in,

[0022] β=ML

[0023] β is the length range of the data packet that can be padded. Φ bytes are added to the end of the data packet P to generate the obfuscated data packet P'.

[0024] Optionally, the packet segmentation mechanism is as follows:

[0025] Let Min be the minimum segment size, Smin and Smax be the minimum and maximum values of the segment respectively, and L be the original length of the data packet P.

[0026] When L >= Min, a random segmentation mechanism is implemented, and Ф is the number of bytes of the data packet P to be segmented.

[0027] Ф = random_int (Smin, Smax)

[0028] Where Smin and Smax are adjustable, and Smax <= MTU.

[0029] Optionally, the false traffic injection mechanism is as follows:

[0030] Let device x be the protection target, F(x) be the traffic of device x, and C(x) be the traffic captured by the attacker about device x.

[0031] Based on the communication traffic between device x, the network service provider ISP and the wireless access point AP, false traffic y is generated, and y is sent together with F(x) in the confusion mode.

[0032] After the attacker obtains the traffic in the form of C(x) = F(x) + y, the events of device x cannot be distinguished by traffic characteristics.

[0033] Optionally, in the process of deconfusion in step 4, for packet padding, the deconfusion module detects and removes redundant data by identifying the marker pre-sent by the sending end; for packet segmentation, the deconfusion module identifies the segmented data packet and reassembles it into the original complete data according to the predetermined order or rule; for false traffic injection, the deconfusion module identifies and processes the false data by identifying the marker pre-sent by the sending end.

[0034] The application provides a user behavior privacy protection method for smart home, which is realized through transmission of a sending end and a receiving end, specifically, a perception link quality module and a dynamic adjustment traffic confusion technology module are deployed in the transmission layer of the sending end in the OSI five-layer model, and a deconfusion module is deployed in the transmission layer of the receiving end; application data of the sending end is transmitted to the transmission layer, the perception link quality module feeds back current network link quality to the dynamic adjustment traffic confusion technology module by monitoring the transmission status of TCP; the dynamic adjustment traffic confusion technology module selects appropriate traffic confusion technology (such as packet segmentation, traffic shaping and false traffic injection) and confusion strength according to the feedback, and performs traffic confusion on the application data, and then encapsulates the data into a data packet and transmits it to the network layer; in the transmission layer of the receiving end, the received data packet is deconfused by the deconfusion module, and the original data before confusion is restored, so that the application layer can obtain accurate application data. The application reduces the user behavior recognition rate to 50% or below without affecting the quality of user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0036] Figure 1 This is a flowchart of the steps of a user behavior privacy protection method for smart homes of the present invention.

[0037] Figure 2 This is a data flow diagram of a user behavior privacy protection method for smart homes in the OSI model of the present invention.

[0038] Figure 3 It is a technical solution diagram of the link quality perception module and the dynamic traffic obfuscation technology module in the method of the present invention.

[0039] Figure 4 It is a schematic diagram of an implementation case of false traffic injection in the method of the present invention.

[0040] Figure 5 Schematic diagram of the deobfuscation principle of the deobfuscation module in the method of the present invention. DETAILED DESCRIPTION

[0041] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0042] The following is an explanation of the abbreviations of English terms used in this invention. If abbreviations are used in the text, please refer to the following explanations:

[0043] Open Systems Interconnection, OSI: Open Systems Interconnection model;

[0044] Transmission Control Protocol, TCP: Transmission Control Protocol;

[0045] Maximum Transmission Unit, MTU: Maximum transmission unit;

[0046] Internet Service Provider, ISP: Internet Service Provider;

[0047] Access Point, AP: wireless access point.

[0048] See also Figure 1 The present invention provides a user behavior privacy protection method for smart home, comprising the following steps:

[0049] S1: The sending end deploys a link quality perception module and a dynamic traffic obfuscation technology module at the transport layer in the five-layer model of Open Systems Interconnection. The receiving end deploys a deobfuscation module at the transport layer.

[0050] S2: The sending end sends application data through the application layer. The link quality perception module monitors the transmission status of the transmission control protocol and feeds back the current network link quality to the dynamic traffic obfuscation technology module.

[0051] S3: The dynamic traffic obfuscation technology module selects the traffic obfuscation technology and obfuscation intensity based on the feedback, performs traffic obfuscation on the application data, and then encapsulates it into a data packet and transmits it to the network layer;

[0052] S4: The receiving end obtains the data packet from the network layer, deobfuscates it through the deobfuscation module, and restores the original data before obfuscation.

[0053] Specifically, the present invention senses link quality and uses a variety of obfuscation technologies in a coordinated manner to reduce the user behavior recognition rate to 50% or below without affecting the user experience quality.

[0054] The main modules include the link quality perception module, the dynamic traffic obfuscation technology module and the deobfuscation module, all of which are deployed in the transport layer of the OSI five-layer model. The data flow of the user behavior privacy protection method for smart homes in the OSI model is shown as follows: Figure 2 shown.

[0055] First, the application data from the sender is passed to the transport layer. The link quality perception module monitors the transmission status of TCP and feeds back the current network link quality to the dynamic traffic obfuscation technology module.

[0056] Secondly, the traffic obfuscation technology module dynamically adjusts the traffic obfuscation technology (such as packet segmentation, traffic shaping, and false traffic injection, etc.) and obfuscation intensity based on the feedback, performs traffic obfuscation on the application data, and then encapsulates it into data packets and transmits it to the network layer.

[0057] Finally, at the transport layer of the receiving end, the received data packets are deobfuscated by the deobfuscation module to restore the original data before obfuscation, so that the application layer can obtain accurate application data.

[0058] The following is further explained in conjunction with specific modules:

[0059] As Figure 3 The technical solution between the link quality perception module and the dynamic adjustment of traffic confusion technology module is shown.

[0060] 1. Link quality perception module

[0061] The link quality perception module is used to timely feedback the current link communication quality, providing a reference for subsequent dynamic adjustment of traffic confusion technology. This module perceives the link quality based on the congestion control mechanism of TCP.

[0062] This module comprehensively evaluates the network link quality by monitoring key parameters of the transport layer TCP connection, such as the sending window size, slow start threshold, round-trip time, etc. By analyzing the dynamic changes of these parameters, this module can accurately detect the congestion state, delay situation and potential bottlenecks of the network, thereby providing strong data support for network optimization and performance tuning. In addition, by monitoring indicators such as the number of retransmissions and packet loss rate of TCP connections, this module can also identify abnormal phenomena in the network, helping to improve the reliability and stability of data transmission.

[0063] 2. Dynamic adjustment of traffic confusion technology module

[0064] The dynamic adjustment of traffic confusion technology module is used to select appropriate traffic confusion technology and confusion strength according to the feedback from the link quality perception module.

[0065] The parameters that need to be adjusted and optimized by this module are as follows:

[0066] 2.1 For data packet padding, the confusion strength refers to the size of β (the smaller β is, the smaller the range of packet padding length, and the weaker the confusion strength). The padding mechanism is as follows:

[0067] Let M be the maximum transmission unit MTU, L be the original length of the data packet P, and Ф be the number of bytes to be added to the data packet P, randomly taking a value between 1 and β.

[0068] Ф = random_int(1, β)

[0069] Where,

[0070] β = M - L

[0071] Ф bytes are added to the end of the data packet P to generate the confused data packet P'.

[0072] 2.2 For data packet segmentation, when Min is the same, the confusion strength is related to Smin (the smaller Smin is, the more fragmented the packet segmentation is, and the stronger the confusion strength is). The segmentation mechanism is as follows:

[0073] Let Min be the minimum segment size, Smin and Smax be the minimum and maximum values ​​of the segment respectively, and L be the original length of the data packet P.

[0074] When L≥Min, a random segmentation mechanism is implemented, and Ф is the number of bytes that the data packet P needs to be segmented.

[0075] Ф=random_int(Smin,Smax)

[0076] Smin and Smax are adjustable, and Smax ≤ MTU.

[0077] 2.3 Regarding false traffic injection, the intensity of obfuscation is related to the injection frequency (the higher the frequency, the more interference and the stronger the obfuscation intensity). The false traffic injection mechanism is as follows:

[0078] Let device x be the protection target, F(x) is the traffic of device x, and C(x) is the traffic about device x captured by the attacker;

[0079] Generate fake traffic y based on the communication traffic between device x, the Internet service provider ISP, and the wireless access point AP. In obfuscation mode, send y together with F(x).

[0080] After the attacker obtains the traffic in the form of C(x)=F(x)+y, he cannot distinguish the events of device x through the traffic characteristics.

[0081] The following is an illustration using a specific implementation case. For example, when device x performs an operation, a fixed traffic pattern is generated. The attacker uses machine learning and other technologies to generate event-level traffic features for device x. By capturing the traffic F(x) of device x and matching the traffic features, the attacker can infer the operation of device x.

[0082] ① When device x performs an operation, the original network traffic is captured, the direction of the data packet is determined, and the constructed false traffic y uses the same destination / source IP, port number, etc. as the traffic of device x. The payload is filled using the packet construction tool and dynamically adjusted.

[0083] ② Send y together with F(x), and the attacker gets the traffic of C(x)=F(x)+y, such as Figure 4 shown.

[0084] ③ Mix the fake traffic y with F(x), making it difficult for the attacker to successfully match it with the traffic characteristics.

[0085] 3. Deobfuscation module

[0086] The deobfuscation module is used at the receiving end to parse and restore the data packets received by the transport layer, and restore the obfuscated data to its original unobfuscated state. The deobfuscation principle of the deobfuscation module is shown as follows: Figure 5As shown.

[0087] The specific de-mixing method is as follows:

[0088] ① For data packet padding, the sending end inserts specific padding data in the data packet to ensure that the data packet reaches the required size or shape. In order to distinguish between padding data and actual data, the sending end uses specific identifiers (such as fixed strings or special byte sequences) to mark these padding data. The receiving end can accurately detect and remove the redundant padding data by identifying these identifiers, thereby recovering the original valid data.

[0089] ② For data packet segmentation, the sending end will segment larger data blocks (exceeding the minimum segment size set by the sending end) into multiple smaller data packets when transmitting data. In the receiving end, the de-mixing module is responsible for identifying these segmented packets and reassembling them into the original complete data according to the predetermined order or rules. The methods for identifying these segmented packets include: using the sequence number in the packet header information to determine the order of the data packets, using the segmentation identifier to identify which data packets belong to the same segmentation process, using the checksum to verify the integrity of the data packets, and using the end marker to indicate the end of data transmission. The combination of these techniques enables the receiving end to effectively identify and reassemble the segmented data packets, ensuring the integrity and correctness of data transmission.

[0090] ③ For false traffic injection, in the sending end, false traffic injection is achieved by inserting fake data packets between real data packets. The de-mixing module in the receiving end needs to identify and filter out these fake false traffic. The fake data packets constructed by the sending end will carry specific markers or formats to distinguish them from normal data packets. The de-mixing module identifies false traffic by detecting these specific markers and discards or ignores them, only retaining and processing valid real data packets. In this way, the receiving end can ensure that the received data is accurate and reliable.

[0091] In summary, compared with the traditional method of using a single mixing technique, the present application dynamically adjusts the traffic mixing technique and mixing strength according to the user's network status. When the user's network status is good, the mixing technique with larger overhead and stronger mixing strength can be selected, and when the user's network status is poor, the mixing technique with smaller overhead and weaker mixing strength can be selected. This makes the mixing scheme more flexible and more conducive to achieving the goal of reducing user behavior recognition rate within the acceptable range of overhead.

[0092] Compared with the prior art, the beneficial effects of the present application are:

[0093] (1) Guarantee user experience quality: the perceived link quality module dynamically monitors network status to ensure that the selection and adjustment of mixing techniques do not affect user experience, especially for low-latency tolerant smart home devices.

[0094] (2) Flexible adaptation to network conditions: Dynamically adjust the traffic obfuscation technology module based on real-time network link quality feedback, select the most appropriate obfuscation technology and intensity, and optimize communication efficiency.

[0095] (3) Easy deployment: The present invention is deployed and implemented only in the transport layer of the OSI five-layer model, and has good versatility and operability.

[0096] The above disclosure is only a preferred embodiment of the present invention, and certainly cannot be used to limit the scope of the rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A user behavior privacy protection method for smart home, characterized by: The following steps are involved: Step 1: The transmitter deploys a link quality perception module and a dynamic traffic obfuscation technology module at the transport layer in the five-layer model of Open Systems Interconnection. The receiver deploys a deobfuscation module at the transport layer. Step 2: The sending end sends application data through the application layer. The link quality perception module monitors the transmission status of the transmission control protocol and feeds back the current network link quality to the dynamic traffic obfuscation technology module; Step 3: The dynamic traffic obfuscation technology module selects the traffic obfuscation technology and obfuscation intensity based on the feedback, performs traffic obfuscation on the application data, and then encapsulates it into a data packet and transmits it to the network layer; Step 4: The receiving end obtains the data packet from the network layer, deobfuscates it through the deobfuscation module, and restores the original data before obfuscation.

2. The user behavior privacy protection method for smart home according to claim 1, characterized in that: The execution process of step 2 includes the following steps: Step 2.1: The sender sends application data to the transport layer through the application layer; Step 2.2: Monitor key parameters of the transport layer TCP connection; Step 2.3: Conduct network link quality assessment; Step 2.4: Perform anomaly detection to identify network anomalies.

3. The user behavior privacy protection method for smart home according to claim 2, characterized in that: The key parameters of the transport layer transmission control protocol connection in step 2 include the sending window size, slow start threshold and round-trip time. The link quality perception module evaluates the network link quality by analyzing the round-trip time, congestion status, delay and bottleneck.

4. The user behavior privacy protection method for smart home according to claim 3, characterized in that: The traffic obfuscation technology in step 3 includes packet padding, packet segmentation, and false traffic injection. Packet padding determines the obfuscation strength by the small / large range of the padding length, packet segmentation determines the obfuscation strength by the simplicity / complexity, and false traffic injection determines the obfuscation strength by the low / high injection frequency.

5. The user behavior privacy protection method for smart home according to claim 4, characterized in that: The padding mechanism for packet padding is as follows: Let M be the maximum transmission unit MTU, L be the original length of data packet P, Φ be the number of bytes to be added to data packet P, and take a random value between 1 and β; Ф=random_int(1,β) in, β=ML β is the length range of the data packet that can be padded. Φ bytes are added to the end of the data packet P to generate the obfuscated data packet P'.

6. The user behavior privacy protection method for smart home according to claim 5, characterized in that: The packet segmentation mechanism is as follows: Let Min be the minimum segment size, Smin and Smax be the minimum and maximum values ​​of the segment respectively, and L be the original length of the data packet P; When L≥Min, a random splitting mechanism is implemented, and Ф is the number of bytes to be split in the data packet P; Ф=random_int(Smin,Smax) Smin and Smax are adjustable, and Smax ≤ MTU.

7. The user behavior privacy protection method for smart home according to claim 6, characterized in that: The fake traffic injection mechanism is as follows: Let device x be the protection target, F(x) is the traffic of device x, and C(x) is the traffic about device x captured by the attacker; Generate fake traffic y based on the communication traffic between device x, the network service provider, and the wireless access point, and send y together with F(x) in obfuscation mode; After the attacker obtains the traffic in the form of C(x)=F(x)+y, he cannot distinguish the events of device x through the traffic characteristics.

8. The user behavior privacy protection method for smart home according to claim 7, characterized in that: During the deobfuscation process in step 4, for data packet padding, the deobfuscation module removes redundant data by identifying the markers pre-sent by the sender; for data packet segmentation, the deobfuscation module identifies the segmented data packets and reassembles them into the original complete data according to a predetermined order or rule; for false traffic injection, the deobfuscation module identifies and processes false data by identifying the markers pre-sent by the sender.

Images